Medical Students in the Emergency Department and Patient Length of Stay

Quantitative assessments of how trainees affect patient care have been limited, especially in the emergency department (ED). A US study by Pitts et al¹ found that supervised resident visits were associated with greater resource use, including longer length of stay (LOS) in the ED. As EDs host more core clerkship courses, less experienced students have become involved in bedside care.² This study examined the association between the presence of medical students in the ED and patient LOS, an established patient-centered outcome and marker of ED performance.³

During a required third-year emergency medicine clerkship at 3 urban, academic EDs associated with the University of Pennsylvania Health System, Philadelphia, students were assigned approximately nine 8- to 12-hour shifts over 3 weeks, during which they were expected to evaluate and follow-up several patients presenting to the ED. The institutional review board at the University of Pennsylvania approved this study and provided a waiver of participant consent.

During the fourth week of each rotation, students participated in an anesthesiology week and were absent from the ED. We examined sequential patient visits from 2000 through 2014, calculating LOS from arrival until ED discharge or admission, and comparing clerkship student presence with student absence from the ED. Summer and winter vacation periods were excluded.

Multivariable generalized linear models included visit-level covariates and dummy variables for all clerkship weeks, and for weeks 1, 2, and 3 individually, and used a γ distribution and clustering by day and hospital. Baseline differences in covariates were assessed for potential confounding factors along with the prevalence of International Classification of Diseases, Ninth Revision (ICD-9), diagnosis codes (using the Benjamini-Hochberg correction for multiple comparisons).

Sensitivity analyses were performed by assessing unadjusted differences using t tests, examining each year and hospital individually, and reassigning visits in which patients left without being seen to varying percentiles of LOS. Two-sided statistical tests with an α level of .05 were performed using Stata version 13 (StataCorp).

More than 1.3 million ED visits were analyzed (Table 1). There were no significant differences among visit covariates, including ICD-9 code prevalence, between clerkship and control weeks. Weekly resident turnover rate was significantly lower during the clerkship weeks compared with the control weeks (mean [SD], 17.8% [16.1%] vs 19.8% [17.5%], respectively), but was not correlated with LOS.

Mean (SD) LOS was 264.7 (253.7) minutes overall; adjusted LOS was 4.6 minutes (95% CI, 2.7-6.6 minutes) longer (P < .001) when clerkship students were present in the ED (Table 2). This was significant across all 3 hospitals and consistent across each of the 3 clerkship weeks. Unadjusted differences for the primary outcome and the sensitivity analysis of left without being seen visits were similar. Subanalysis of each year at each site showed that LOS was either longer when students were present or not significantly different from the control weeks.

Our findings show an increase in LOS of approximately 5 minutes associated with the presence of medical students in the ED, which was statistically significant but likely too small to be of clinical relevance (equivalent to 2% of 1 SD in LOS). This conclusion was robust for all sensitivity analyses and persisted across 3 different hospitals with distinct teaching models, patient populations, and workflow. Prior studies have had conflicting results and only demonstrated longer LOS for the select patients examined by students directly.^4-6

An important limitation is the absence of visit-level information on student involvement with patients. Without this, we were able to examine only the association of the clerkship with aggregate overall LOS. In addition, indirect effects, such as attending physicians spending extra time documenting after shifts as a consequence of teaching students during clinical time, were not assessed, and this study took place at 3 hospitals associated with a single medical school. Future studies should assess different student experiences and other patient-centered or financial outcomes.

Corresponding Author: Kevin R. Scott, MD, Department of Emergency Medicine, 3400 Spruce St, Ground Ravdin, Philadelphia, PA 19104 (kevin.scott@uphs.upenn.edu).

Author Contributions: Mr Ioannides had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Ioannides, Hennessy, Scott.

Acquisition, analysis, or interpretation of data: Ioannides, Shofer, Small, Hennessy, Abella, Scott.

Drafting of the manuscript: Ioannides, Mamtani, Scott.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Ioannides, Shofer, Small.

Administrative, technical, or material support: Small, Abella, Scott.

Study supervision: Mamtani, Abella, Scott.

Conflict of Interest Disclosures: The authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Dr Abella reported receiving grants from the National Heart, Lung, and Blood Institute, Philips Healthcare, Medtronic Foundation, and Stryker Medical; and receiving other (honoraria, stock ownership, serving on advisory boards, or serving as a consultant) from Velomedix Corp, HeartSine Corp, Ikaria Inc, Resuscor, Bard, and Stryker Medical. No other disclosures were reported.

Funding/Support: The Department of Emergency Medicine, University of Pennsylvania, provided funding for this study.

Role of the Funder/Sponsor: The Department of Emergency Medicine, University of Pennsylvania, had no role in the design and conduct of the study; had a role in the collection, management, analysis, and interpretation of the data; had no role in the preparation, review, or approval of the manuscript; and had no role in the decision to submit the manuscript for publication.

Previous Presentations: A preliminary version of these data were presented in part at a pharmacoepidemiology rounds on April 17, 2015, at the University of Pennsylvania and at an internal research-in-progress meeting on April 29, 2015.

Additional Contributions: We thank the following individuals at the University of Pennsylvania for their contributions without compensation: Carmen Farrow, MS, MBA, for special data analysis and technical assistance with EMR databases; Ari B. Friedman, PhD, for analysis and operational feedback; Jennifer R. Kogan, MD, for advising and providing feedback; and Elizabeth D. Lowenthal, MD, MSCE, and Iris M. Reyes, MD, for advising and providing assistance with the protocol.